bacow/chatlog
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Support Media Message (#9)

Browse Source
This commit is contained in:
Sarv
2025-03-28 16:48:49 +08:00
committed by GitHub
parent 98f41454fb
commit f31953c42b
24 changed files with 1428 additions and 136 deletions
Download Patch File Download Diff File Expand all files Collapse all files

105
internal/wechatdb/datasource/darwinv3/datasource.go
View File

@@ -6,7 +6,6 @@ import (
"database/sql"
"encoding/hex"
"fmt"
"log"
"strings"
"time"
@@ -14,6 +13,7 @@ import (
"github.com/sjzar/chatlog/pkg/util"
_ "github.com/mattn/go-sqlite3"
log "github.com/sirupsen/logrus"
)
const (
@@ -21,6 +21,7 @@ const (
ContactFilePattern = "^wccontact_new2\\.db$"
ChatRoomFilePattern = "^group_new\\.db$"
SessionFilePattern = "^session_new\\.db$"
MediaFilePattern = "^hldata\\.db$"
)
type DataSource struct {
@@ -29,6 +30,7 @@ type DataSource struct {
contactDb *sql.DB
chatRoomDb *sql.DB
sessionDb *sql.DB
mediaDb *sql.DB
talkerDBMap map[string]*sql.DB
user2DisplayName map[string]string
@@ -54,6 +56,9 @@ func New(path string) (*DataSource, error) {
if err := ds.initSessionDb(path); err != nil {
return nil, fmt.Errorf("初始化会话数据库失败: %w", err)
}
if err := ds.initMediaDb(path); err != nil {
return nil, fmt.Errorf("初始化会话数据库失败: %w", err)
}
return ds, nil
}
@@ -138,7 +143,7 @@ func (ds *DataSource) initChatRoomDb(path string) error {
return fmt.Errorf("连接群聊数据库失败: %w", err)
}
rows, err := ds.chatRoomDb.Query("SELECT m_nsUsrName, nickname FROM GroupMember")
rows, err := ds.chatRoomDb.Query("SELECT m_nsUsrName, IFNULL(nickname,\"\") FROM GroupMember")
if err != nil {
log.Printf("警告: 获取群聊成员失败: %v", err)
return nil
@@ -173,6 +178,21 @@ func (ds *DataSource) initSessionDb(path string) error {
return nil
}
func (ds *DataSource) initMediaDb(path string) error {
files, err := util.FindFilesWithPatterns(path, MediaFilePattern, true)
if err != nil {
return fmt.Errorf("查找媒体数据库文件失败: %w", err)
}
if len(files) == 0 {
return fmt.Errorf("未找到媒体数据库文件: %s", path)
}
ds.mediaDb, err = sql.Open("sqlite3", files[0])
if err != nil {
return fmt.Errorf("连接媒体数据库失败: %w", err)
}
return nil
}
// GetMessages 实现获取消息的方法
func (ds *DataSource) GetMessages(ctx context.Context, startTime, endTime time.Time, talker string, limit, offset int) ([]*model.Message, error) {
// 在 darwinv3 中，每个联系人/群聊的消息存储在单独的表中，表名为 Chat_md5(talker)
@@ -191,7 +211,7 @@ func (ds *DataSource) GetMessages(ctx context.Context, startTime, endTime time.T
// 构建查询条件
query := fmt.Sprintf(`
SELECT msgCreateTime, msgContent, messageType, mesDes, msgSource, CompressContent, ConBlob
SELECT msgCreateTime, msgContent, messageType, mesDes
FROM %s
WHERE msgCreateTime >= ? AND msgCreateTime <= ?
ORDER BY msgCreateTime ASC
@@ -216,15 +236,11 @@ func (ds *DataSource) GetMessages(ctx context.Context, startTime, endTime time.T
messages := []*model.Message{}
for rows.Next() {
var msg model.MessageDarwinV3
var compressContent, conBlob []byte
err := rows.Scan(
&msg.MesCreateTime,
&msg.MesContent,
&msg.MesType,
&msg.MsgCreateTime,
&msg.MsgContent,
&msg.MessageType,
&msg.MesDes,
&msg.MesSource,
&compressContent,
&conBlob,
)
if err != nil {
log.Printf("警告: 扫描消息行失败: %v", err)
@@ -260,13 +276,13 @@ func (ds *DataSource) GetContacts(ctx context.Context, key string, limit, offset
if key != "" {
// 按照关键字查询
query = `SELECT IFNULL(m_nsUsrName,""), nickname, IFNULL(m_nsRemark,""), m_uiSex, IFNULL(m_nsAliasName,"")
query = `SELECT IFNULL(m_nsUsrName,""), IFNULL(nickname,""), IFNULL(m_nsRemark,""), m_uiSex, IFNULL(m_nsAliasName,"")
FROM WCContact
WHERE m_nsUsrName = ? OR nickname = ? OR m_nsRemark = ? OR m_nsAliasName = ?`
args = []interface{}{key, key, key, key}
} else {
// 查询所有联系人
query = `SELECT IFNULL(m_nsUsrName,""), nickname, IFNULL(m_nsRemark,""), m_uiSex, IFNULL(m_nsAliasName,"")
query = `SELECT IFNULL(m_nsUsrName,""), IFNULL(nickname,""), IFNULL(m_nsRemark,""), m_uiSex, IFNULL(m_nsAliasName,"")
FROM WCContact`
}
@@ -314,13 +330,13 @@ func (ds *DataSource) GetChatRooms(ctx context.Context, key string, limit, offse
if key != "" {
// 按照关键字查询
query = `SELECT IFNULL(m_nsUsrName,""), nickname, IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
query = `SELECT IFNULL(m_nsUsrName,""), IFNULL(nickname,""), IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
FROM GroupContact
WHERE m_nsUsrName = ? OR nickname = ? OR m_nsRemark = ?`
args = []interface{}{key, key, key}
} else {
// 查询所有群聊
query = `SELECT IFNULL(m_nsUsrName,""), nickname, IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
query = `SELECT IFNULL(m_nsUsrName,""), IFNULL(nickname,""), IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
FROM GroupContact`
}
@@ -364,7 +380,7 @@ func (ds *DataSource) GetChatRooms(ctx context.Context, key string, limit, offse
if err == nil && len(contacts) > 0 && strings.HasSuffix(contacts[0].UserName, "@chatroom") {
// 再次尝试通过用户名查找群聊
rows, err := ds.chatRoomDb.QueryContext(ctx,
`SELECT IFNULL(m_nsUsrName,""), nickname, IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
`SELECT IFNULL(m_nsUsrName,""), IFNULL(nickname,""), IFNULL(m_nsRemark,""), IFNULL(m_nsChatRoomMemList,""), IFNULL(m_nsChatRoomAdminList,"")
FROM GroupContact
WHERE m_nsUsrName = ?`,
contacts[0].UserName)
@@ -470,6 +486,58 @@ func (ds *DataSource) GetSessions(ctx context.Context, key string, limit, offset
return sessions, nil
}
func (ds *DataSource) GetMedia(ctx context.Context, _type string, key string) (*model.Media, error) {
if key == "" {
return nil, fmt.Errorf("key 不能为空")
}
query := `SELECT
r.mediaMd5,
r.mediaSize,
r.inodeNumber,
r.modifyTime,
d.relativePath,
d.fileName
FROM
HlinkMediaRecord r
JOIN
HlinkMediaDetail d ON r.inodeNumber = d.inodeNumber
WHERE
r.mediaMd5 = ?`
args := []interface{}{key}
// 执行查询
rows, err := ds.mediaDb.QueryContext(ctx, query, args...)
if err != nil {
return nil, fmt.Errorf("查询媒体失败: %w", err)
}
defer rows.Close()
var media *model.Media
for rows.Next() {
var mediaDarwinV3 model.MediaDarwinV3
err := rows.Scan(
&mediaDarwinV3.MediaMd5,
&mediaDarwinV3.MediaSize,
&mediaDarwinV3.InodeNumber,
&mediaDarwinV3.ModifyTime,
&mediaDarwinV3.RelativePath,
&mediaDarwinV3.FileName,
)
if err != nil {
return nil, fmt.Errorf("扫描会话行失败: %w", err)
}
// 包装成通用模型
media = mediaDarwinV3.Wrap()
}
if media == nil {
return nil, fmt.Errorf("未找到媒体 %s", key)
}
return media, nil
}
// Close 实现关闭数据库连接的方法
func (ds *DataSource) Close() error {
var errs []error
@@ -502,6 +570,13 @@ func (ds *DataSource) Close() error {
}
}
// 关闭媒体数据库连接
if ds.mediaDb != nil {
if err := ds.mediaDb.Close(); err != nil {
errs = append(errs, fmt.Errorf("关闭媒体数据库失败: %w", err))
}
}
if len(errs) > 0 {
return fmt.Errorf("关闭数据库连接时发生错误: %v", errs)
}

3
internal/wechatdb/datasource/datasource.go
View File

@@ -30,6 +30,9 @@ type DataSource interface {
// 最近会话
GetSessions(ctx context.Context, key string, limit, offset int) ([]*model.Session, error)
// 媒体
GetMedia(ctx context.Context, _type string, key string) (*model.Media, error)
Close() error
}

101
internal/wechatdb/datasource/v4/datasource.go
View File

@@ -21,6 +21,7 @@ const (
MessageFilePattern = "^message_([0-9]?[0-9])?\\.db$"
ContactFilePattern = "^contact\\.db$"
SessionFilePattern = "^session\\.db$"
MediaFilePattern = "^hardlink\\.db$"
)
// MessageDBInfo 存储消息数据库的信息
@@ -36,6 +37,7 @@ type DataSource struct {
messageDbs map[string]*sql.DB
contactDb *sql.DB
sessionDb *sql.DB
mediaDb *sql.DB
// 消息数据库信息
messageFiles []MessageDBInfo
@@ -57,6 +59,9 @@ func New(path string) (*DataSource, error) {
if err := ds.initSessionDb(path); err != nil {
return nil, fmt.Errorf("初始化会话数据库失败: %w", err)
}
if err := ds.initMediaDb(path); err != nil {
return nil, fmt.Errorf("初始化媒体数据库失败: %w", err)
}
return ds, nil
}
@@ -175,6 +180,21 @@ func (ds *DataSource) initSessionDb(path string) error {
return nil
}
func (ds *DataSource) initMediaDb(path string) error {
files, err := util.FindFilesWithPatterns(path, MediaFilePattern, true)
if err != nil {
return fmt.Errorf("查找媒体数据库文件失败: %w", err)
}
if len(files) == 0 {
return fmt.Errorf("未找到媒体数据库文件: %s", path)
}
ds.mediaDb, err = sql.Open("sqlite3", files[0])
if err != nil {
return fmt.Errorf("连接媒体数据库失败: %w", err)
}
return nil
}
// getDBInfosForTimeRange 获取时间范围内的数据库信息
func (ds *DataSource) getDBInfosForTimeRange(startTime, endTime time.Time) []MessageDBInfo {
var dbs []MessageDBInfo
@@ -602,6 +622,81 @@ func (ds *DataSource) GetSessions(ctx context.Context, key string, limit, offset
return sessions, nil
}
func (ds *DataSource) GetMedia(ctx context.Context, _type string, key string) (*model.Media, error) {
if key == "" {
return nil, fmt.Errorf("key 不能为空")
}
if len(key) != 32 {
return nil, fmt.Errorf("key 长度必须为 32")
}
var table string
switch _type {
case "image":
table = "image_hardlink_info_v3"
case "video":
table = "video_hardlink_info_v3"
case "file":
table = "file_hardlink_info_v3"
default:
return nil, fmt.Errorf("不支持的媒体类型: %s", _type)
}
query := fmt.Sprintf(`
SELECT
f.md5,
f.file_name,
f.file_size,
f.modify_time,
IFNULL(d1.username,""),
IFNULL(d2.username,"")
FROM
%s f
LEFT JOIN
dir2id d1 ON d1.rowid = f.dir1
LEFT JOIN
dir2id d2 ON d2.rowid = f.dir2
`, table)
query += " WHERE f.md5 = ? OR f.file_name LIKE ? || '%'"
args := []interface{}{key, key}
rows, err := ds.mediaDb.QueryContext(ctx, query, args...)
if err != nil {
return nil, fmt.Errorf("查询媒体失败: %w", err)
}
defer rows.Close()
var media *model.Media
for rows.Next() {
var mediaV4 model.MediaV4
err := rows.Scan(
&mediaV4.Key,
&mediaV4.Name,
&mediaV4.Size,
&mediaV4.ModifyTime,
&mediaV4.Dir1,
&mediaV4.Dir2,
)
if err != nil {
return nil, fmt.Errorf("扫描会话行失败: %w", err)
}
mediaV4.Type = _type
media = mediaV4.Wrap()
// 跳过缩略图
if _type == "image" && !strings.Contains(media.Name, "_t") {
break
}
}
if media == nil {
return nil, fmt.Errorf("未找到媒体 %s", key)
}
return media, nil
}
func (ds *DataSource) Close() error {
var errs []error
@@ -626,6 +721,12 @@ func (ds *DataSource) Close() error {
}
}
if ds.mediaDb != nil {
if err := ds.mediaDb.Close(); err != nil {
errs = append(errs, fmt.Errorf("关闭媒体数据库失败: %w", err))
}
}
if len(errs) > 0 {
return fmt.Errorf("关闭数据库连接时发生错误: %v", errs)
}

153
internal/wechatdb/datasource/windowsv3/datasource.go
View File

@@ -3,6 +3,7 @@ package windowsv3
import (
"context"
"database/sql"
"encoding/hex"
"fmt"
"log"
"sort"
@@ -18,6 +19,9 @@ import (
const (
MessageFilePattern = "^MSG([0-9]?[0-9])?\\.db$"
ContactFilePattern = "^MicroMsg.db$"
ImageFilePattern = "^HardLinkImage\\.db$"
VideoFilePattern = "^HardLinkVideo\\.db$"
FileFilePattern = "^HardLinkFile\\.db$"
)
// MessageDBInfo 保存消息数据库的信息
@@ -37,6 +41,10 @@ type DataSource struct {
// 联系人数据库
contactDbFile string
contactDb *sql.DB
imageDb *sql.DB
videoDb *sql.DB
fileDb *sql.DB
}
// New 创建一个新的 WindowsV3DataSource
@@ -56,6 +64,10 @@ func New(path string) (*DataSource, error) {
return nil, fmt.Errorf("初始化联系人数据库失败: %w", err)
}
if err := ds.initMediaDb(path); err != nil {
return nil, fmt.Errorf("初始化多媒体数据库失败: %w", err)
}
return ds, nil
}
@@ -178,6 +190,53 @@ func (ds *DataSource) initContactDb(path string) error {
return nil
}
// initContactDb 初始化联系人数据库
func (ds *DataSource) initMediaDb(path string) error {
files, err := util.FindFilesWithPatterns(path, ImageFilePattern, true)
if err != nil {
return fmt.Errorf("查找图片数据库文件失败: %w", err)
}
if len(files) == 0 {
return fmt.Errorf("未找到图片数据库文件: %s", path)
}
ds.imageDb, err = sql.Open("sqlite3", files[0])
if err != nil {
return fmt.Errorf("连接图片数据库失败: %w", err)
}
files, err = util.FindFilesWithPatterns(path, VideoFilePattern, true)
if err != nil {
return fmt.Errorf("查找视频数据库文件失败: %w", err)
}
if len(files) == 0 {
return fmt.Errorf("未找到视频数据库文件: %s", path)
}
ds.videoDb, err = sql.Open("sqlite3", files[0])
if err != nil {
return fmt.Errorf("连接视频数据库失败: %w", err)
}
files, err = util.FindFilesWithPatterns(path, FileFilePattern, true)
if err != nil {
return fmt.Errorf("查找文件数据库文件失败: %w", err)
}
if len(files) == 0 {
return fmt.Errorf("未找到文件数据库文件: %s", path)
}
ds.fileDb, err = sql.Open("sqlite3", files[0])
if err != nil {
return fmt.Errorf("连接文件数据库失败: %w", err)
}
return nil
}
// getDBInfosForTimeRange 获取时间范围内的数据库信息
func (ds *DataSource) getDBInfosForTimeRange(startTime, endTime time.Time) []MessageDBInfo {
var dbs []MessageDBInfo
@@ -589,6 +648,84 @@ func (ds *DataSource) GetSessions(ctx context.Context, key string, limit, offset
return sessions, nil
}
func (ds *DataSource) GetMedia(ctx context.Context, _type string, key string) (*model.Media, error) {
if key == "" {
return nil, fmt.Errorf("key 不能为空")
}
md5key, err := hex.DecodeString(key)
if err != nil {
return nil, fmt.Errorf("解析 key 失败: %w", err)
}
var db *sql.DB
var table1, table2 string
switch _type {
case "image":
db = ds.imageDb
table1 = "HardLinkImageAttribute"
table2 = "HardLinkImageID"
case "video":
db = ds.videoDb
table1 = "HardLinkVideoAttribute"
table2 = "HardLinkVideoID"
case "file":
db = ds.fileDb
table1 = "HardLinkFileAttribute"
table2 = "HardLinkFileID"
default:
return nil, fmt.Errorf("不支持的媒体类型: %s", _type)
}
query := fmt.Sprintf(`
SELECT
a.FileName,
a.ModifyTime,
IFNULL(d1.Dir,"") AS Dir1,
IFNULL(d2.Dir,"") AS Dir2
FROM
%s a
LEFT JOIN
%s d1 ON a.DirID1 = d1.DirId
LEFT JOIN
%s d2 ON a.DirID2 = d2.DirId
WHERE
a.Md5 = ?
`, table1, table2, table2)
args := []interface{}{md5key}
rows, err := db.QueryContext(ctx, query, args...)
if err != nil {
return nil, fmt.Errorf("查询媒体失败: %w", err)
}
defer rows.Close()
var media *model.Media
for rows.Next() {
var mediaV3 model.MediaV3
err := rows.Scan(
&mediaV3.Name,
&mediaV3.ModifyTime,
&mediaV3.Dir1,
&mediaV3.Dir2,
)
if err != nil {
return nil, fmt.Errorf("扫描会话行失败: %w", err)
}
mediaV3.Type = _type
mediaV3.Key = key
media = mediaV3.Wrap()
}
if media == nil {
return nil, fmt.Errorf("未找到媒体 %s", key)
}
return media, nil
}
// Close 实现 DataSource 接口的 Close 方法
func (ds *DataSource) Close() error {
var errs []error
@@ -607,6 +744,22 @@ func (ds *DataSource) Close() error {
}
}
if ds.imageDb != nil {
if err := ds.imageDb.Close(); err != nil {
errs = append(errs, fmt.Errorf("关闭图片数据库失败: %w", err))
}
}
if ds.videoDb != nil {
if err := ds.videoDb.Close(); err != nil {
errs = append(errs, fmt.Errorf("关闭视频数据库失败: %w", err))
}
}
if ds.fileDb != nil {
if err := ds.fileDb.Close(); err != nil {
errs = append(errs, fmt.Errorf("关闭文件数据库失败: %w", err))
}
}
if len(errs) > 0 {
return fmt.Errorf("关闭数据库连接时发生错误: %v", errs)
}